Large-Scale Mapping of Maize Plant Density Using Multi-Temporal Optical and Radar Data: Models, Potential and...

Area of Science:

• Agricultural Science
• Remote Sensing
• Data Fusion

Background:

• Accurate crop density estimation is vital for agricultural resource management.
• Existing methods struggle with data acquisition and optical-radar data inconsistencies.
• Challenges hinder effective large-scale crop monitoring and precision agriculture.

Purpose of the Study:

• To develop a novel approach for maize density estimation by integrating optical and radar data.
• To address data inconsistencies and improve model usability for agricultural resource management.
• To enhance the accuracy of crop density estimation across diverse growth stages.

Main Methods:

• A unique mapping strategy combining data selection, feature extraction, and optimization was employed.
• Machine learning was used to identify critical features and optimal combinations for maize density.
• A multi-temporal model was developed integrating optical and radar data for enhanced estimation.

Main Results:

• The integrated approach significantly improved maize density estimation accuracy (R² = 0.602, RMSE = 0.094).
• Enhanced accuracy was observed during key growth stages: leaf development, stem elongation, and tasseling.
• Successful maize density maps were generated for demonstration counties, outperforming single-temporal models.

Conclusions:

• The novel integrated optical-radar data approach advances large-scale maize density estimation.
• This method offers a foundation for optimizing agricultural resource management and precision agriculture.
• The approach has potential for expansion to other regions and crop types.